LightRAG
Graph-based retrieval-augmented generation library with a web UI
What it is
LightRAG is a Python framework that builds a knowledge graph from your documents and combines graph and vector retrieval to answer questions with an LLM. It supports storage backends such as PostgreSQL, Neo4j, MongoDB and OpenSearch, parses PDFs and Office files through MinerU or Docling, and includes a server with a web interface for inserting, querying and visualizing data. Developers building question-answering systems over their own document collections use it as a lighter alternative to heavier GraphRAG pipelines, with RAGAS evaluation and Langfuse tracing built in.